pith. sign in

arxiv: 2606.24899 · v1 · pith:WWI2VPQBnew · submitted 2026-06-12 · 💻 cs.LG · cs.AI· quant-ph

From Meta Idea to Advanced Mathematical Discovery -- Human-AI Co-Discovery of Sign-Embedding Quantum Algorithms

Yanqiao Wang , Jin-Peng Liu , Peng Li , Yang Liu This is my paper

Pith reviewed 2026-06-27 05:05 UTC · model grok-4.3

classification 💻 cs.LG cs.AIquant-ph
keywords human-AI co-discoverysign-embeddingquantum algorithmsmatrix functionsrational approximationsign functionquantum linear algebramatrix equations
0
0 comments X

The pith

Human judgment gates AI expansion of a rational-approximation intuition into sign-embedding quantum algorithms.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper traces how a human intuition that rational approximation works well for jump-type functions like the sign function was turned into concrete research routes and a sign-embedding framework for quantum algorithms on matrix equations and functions. AI assistance helped expand the initial idea, compare formulations, link a matrix-sign identity to wider classes of problems, and draft supporting calculations. All decisive choices about which routes to keep or drop stayed with the human researchers. A reader would care because the account models how AI can support the earliest, least-structured phase of mathematical work inside a human-controlled process rather than by solving problems that are already fixed.

Core claim

The central claim is that human-AI co-discovery workflows are most valuable not as standalone theorem provers but as research partners for problem formation, connection discovery, derivation, and skeptical review inside a human-gated research loop, as shown by the development of sign-embedding quantum algorithms from an initial intuition about rational approximation for the sign function.

What carries the argument

sign embedding, the framework that uses the sign function to connect a known matrix-sign identity to wider classes of matrix equations and matrix functions.

Load-bearing premise

The paper's narrative of which contributions came from the AI and which remained with human judgment accurately reflects the actual sequence of interactions.

What would settle it

A complete record of all AI-human exchanges during the project that shows either the AI never expanded the candidate routes or that the human rejected the invalid Cayley-trapezoidal route before any AI input would falsify the described benefit of the co-discovery workflow.

Figures

Figures reproduced from arXiv: 2606.24899 by Jin-Peng Liu, Peng Li, Yang Liu, Yanqiao Wang.

Figure 1
Figure 1. Figure 1: From a fixed-target proof pipeline to a human-gated research-loop workflow. The workflow studied in this report begins before a theorem is fixed: AI helps expand a human-originated idea into candidate routes and proof-verification loops, while human judgment controls route selection and final audit. become, which route is promising, what assumptions are natural, and what final result would count as a meani… view at source ↗
Figure 2
Figure 2. Figure 2: Human-gated AI research workflow. Unit Lead role Output of the unit Input: human meta idea Human Rational approximation, especially for jump-type functions such as the sign function, may be useful as a design principle for quantum algorithms. Stage 1: divergent route expansion Human-AI interaction A route map of candidate programs, including shifted resolvents, projectors, matrix equations, sign methods, n… view at source ↗
Figure 3
Figure 3. Figure 3: Section-level development map of the 84-page sign-embedding paper [9]. The vertical strip represents the manuscript from front matter to appendices, with section heights roughly proportional to page spans. Colors indicate the dominant development mode of each portion. The map is descriptive rather than quantitative: it is not intended to assign scientific credit, difficulty, or novelty. 3.3 Map of AIM’s Ro… view at source ↗
read the original abstract

AI-assisted mathematics is often evaluated on solving predefined problems. In practice, however, many important advances begin earlier, when a vague research intuition is transformed into a concrete problem, a promising route, and a theorem family worth proving. This report studies that stage through a case study that led to sign-embedding quantum algorithms for matrix equations and matrix functions, foundational primitives in quantum linear algebra and operator-output quantum algorithms. The project began with a human-originated intuition that rational approximation is especially effective for jump-type functions such as the sign function, and might therefore serve as a design principle for quantum algorithms. Rather than merely assisting after the problem was fixed, AI-assisted exploration, including workflows later integrated into the agentic AI-mathematician system AIM, played a key role in expanding this intuition into a route map, comparing candidate formulations, and converging toward sign embedding as the central framework. AIM then helped connect a known matrix-sign identity to wider classes of matrix equations and matrix functions, and drafted proof and complexity calculations. The decisive scientific judgments remained human: selecting which human-AI-expanded routes were worth pursuing, rejecting a Cayley-trapezoidal approximation when its validity required a hidden condition, and refining the Sylvester implementation from a coarse quadratic-gap query route to the final factorized and scaled analysis. The report argues that human-AI co-discovery workflows, with systems such as AIM as important components, are most valuable not as standalone theorem provers, but as research partners for problem formation, connection discovery, derivation, and skeptical review inside a human-gated research loop.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 0 minor

Summary. The paper claims that human-AI co-discovery workflows using systems such as AIM are most valuable for problem formation, connection discovery, derivation, and skeptical review inside a human-gated research loop, as illustrated by a case study in which a human intuition about rational approximations for the sign function was expanded into the sign-embedding framework for quantum algorithms on matrix equations and matrix functions.

Significance. If the attributions in the narrative hold, the work would provide a concrete example of AI assistance at the pre-theorem stage of research and credit the human role in route selection and validation. The absence of any independent verification of the AI interactions, however, prevents assessment of whether the claimed division of labor is accurate.

major comments (2)
  1. [Abstract] Abstract: The central claim that AIM 'helped connect a known matrix-sign identity to wider classes... and drafted proof and complexity calculations' rests entirely on the authors' self-reported narrative with no referenced interaction logs, prompt histories, or AIM output artifacts, so the effectiveness of the co-discovery process cannot be externally verified.
  2. [Abstract] Abstract: The argument that human judgments alone selected routes and rejected the Cayley-trapezoidal approximation is defined in terms of the same internal sequence the authors recount, creating a circularity that prevents separation of the process claim from the outcome.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive feedback on verifiability and potential circularity in the case-study narrative. We respond to each major comment below. The manuscript is a reflective report on a human-gated workflow rather than a controlled experiment, which limits the forms of external verification that can be supplied.

read point-by-point responses

  1. Referee: [Abstract] Abstract: The central claim that AIM 'helped connect a known matrix-sign identity to wider classes... and drafted proof and complexity calculations' rests entirely on the authors' self-reported narrative with no referenced interaction logs, prompt histories, or AIM output artifacts, so the effectiveness of the co-discovery process cannot be externally verified.

    Authors: We agree that the paper presents a narrative case study rather than timestamped logs or artifacts. The mathematical contributions (sign-embedding algorithms for matrix equations and functions) are stated in full and can be checked independently of the discovery story. Because the interactions occurred over an extended period and were not archived in a machine-readable, shareable format at the time, we cannot retroactively supply the requested logs. We will revise the abstract and add a short methods paragraph clarifying that the account is based on participant recollection and working notes, while emphasizing that the final theorems stand on their own. revision: partial

  2. Referee: [Abstract] Abstract: The argument that human judgments alone selected routes and rejected the Cayley-trapezoidal approximation is defined in terms of the same internal sequence the authors recount, creating a circularity that prevents separation of the process claim from the outcome.

    Authors: The manuscript separates the verifiable mathematical outcome (the sign-embedding framework and its complexity bounds) from the process description. The rejection of the Cayley-trapezoidal route is justified inside the paper by the explicit discovery of a hidden validity condition that renders the approximation unusable for the target matrix classes; this is a mathematical observation, not merely an assertion of human preference. We will add one sentence to the abstract stating that the process narrative is offered as an illustration of human oversight within the loop, while the algorithmic results are presented for independent evaluation. revision: partial

standing simulated objections not resolved
  • Absence of archived interaction logs, prompt histories, or AIM output artifacts from the original research sessions, which prevents any form of external verification of the specific attributions made to the AI system.

Circularity Check

0 steps flagged

No circularity detected; narrative case study lacks derivation chain or load-bearing reductions

full rationale

The paper is a descriptive narrative case study of a human-AI research process for developing sign-embedding quantum algorithms. It contains no mathematical derivations, equations, fitted parameters, predictions, uniqueness theorems, or ansatzes that could reduce to inputs by construction. The enumerated circularity patterns (self-definitional, fitted_input_called_prediction, self_citation_load_bearing, etc.) do not apply, as there are no such elements to inspect. The central claim about workflow value is supported by the recounted sequence itself, but this is a process report rather than a theorem whose result is forced by definition or self-citation. Per rules, this warrants score 0 as an honest non-finding for a self-contained narrative without the specified reductions.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 1 invented entities

The paper introduces sign-embedding as the converged framework but supplies no external evidence or derivation independent of the narrated process; no free parameters or standard axioms are explicitly listed beyond the background of quantum linear algebra.

invented entities (1)
  • sign-embedding framework no independent evidence
    purpose: Central organizing structure for quantum algorithms targeting matrix equations and functions via sign-function approximations
    Presented as the outcome of the human-AI route convergence; no independent falsifiable prediction or external validation is supplied in the abstract.

pith-pipeline@v0.9.1-grok · 5829 in / 1278 out tokens · 22255 ms · 2026-06-27T05:05:32.649444+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

73 extracted references · 28 canonical work pages · 2 internal anchors

  1. [1]

    BernardinoRomera-Paredes,MohammadaminBarekatain,AlexanderNovikov,MatejBalog,M.PawanKumar, Emilien Dupont, Francisco J. R. Ruiz, Jordan S. Ellenberg, Pengming Wang, Omar Fawzi, Pushmeet Kohli, and Alhussein Fawzi. Mathematical discoveries from program search with large language models.Nature, 625 (7995):468–475,2024. doi: 10.1038/s41586-023-06924-6. URLhtt...

    work page doi:10.1038/s41586-023-06924-6 2024

  2. [2]

    Alhussein Fawzi, Matej Balog, Aja Huang, Thomas Hubert, Bernardino Romera-Paredes, Mohammadamin Barekatain, Alexander Novikov, Francisco J. R. Ruiz, Julian Schrittwieser, Grzegorz Swirszcz, David Silver, Demis Hassabis, and Pushmeet Kohli. Discovering faster matrix multiplication algorithms with reinforcement learning.Nature, 610:47–53, 2022. doi: 10.1038...

    work page doi:10.1038/s41586-022-05172-4 2022

  3. [3]

    URL https://openai.com/index/model-disproves-discrete-geometry-conjecture/

    OpenAI.AnOpenAImodelhasdisprovedacentralconjectureindiscretegeometry.OpenAIResearchMilestone, May 2026. URL https://openai.com/index/model-disproves-discrete-geometry-conjecture/. Published May 20, 2026

    2026

  4. [4]

    Nature625, 476–482 (2024).https://doi.org/ 10.1038/s41586-023-06747-5

    Trieu H. Trinh, Yuhuai Wu, Quoc V. Le, He He, and Thang Luong. Solving olympiad geometry without human demonstrations.Nature, 625(7995):476–482, 2024. doi: 10.1038/s41586-023-06747-5. URL https://doi.org/10.1038/s41586-023-06747-5

    work page doi:10.1038/s41586-023-06747-5 2024

  5. [5]

    Olympiad-level formal mathematical reasoning with reinforcement learning

    Thomas Hubert, Rishi Mehta, Laurent Sartran, Miklós Z. Horváth, Goran Žužić, Eric Wieser, Aja Huang, Julian Schrittwieser, Yannick Schroecker, Hussain Masoom, et al. Olympiad-level formal mathematical reasoning with reinforcement learning.Nature, 651:607–613, 2026. doi: 10.1038/s41586-025-09833-y. URL https://doi.org/10.1038/s41586-025-09833-y. Published ...

    work page doi:10.1038/s41586-025-09833-y 2026

  6. [6]

    AI mathematician: Towards fully automated frontier mathematical research.arXiv preprint arXiv:2505.22451, 2025

    Yuanhang Liu, Yanxing Huang, Yanqiao Wang, Peng Li, and Yang Liu. AI mathematician: Towards fully automated frontier mathematical research.arXiv preprint arXiv:2505.22451, 2025. URL https: //arxiv.org/abs/2505.22451

    arXiv 2025

  7. [7]

    AI mathematician as a partner in advancing mathematical discovery—a case study in homogenization theory.arXiv preprint arXiv:2510.26380, 2025

    Yuanhang Liu, Beichen Wang, Peng Li, and Yang Liu. AI mathematician as a partner in advancing mathematical discovery—a case study in homogenization theory.arXiv preprint arXiv:2510.26380, 2025. URL https://arxiv.org/abs/2510.26380

    arXiv 2025

  8. [8]

    Trinh, Garrett Bingham, Dawsen Hwang, Yuri Chervonyi, Junehyuk Jung, Joonkyung Lee, Carlo Pagano, Sang-hyun Kim, Federico Pasqualotto, Sergei Gukov, Jonathan N

    Tony Feng, Trieu H. Trinh, Garrett Bingham, Dawsen Hwang, Yuri Chervonyi, Junehyuk Jung, Joonkyung Lee, Carlo Pagano, Sang-hyun Kim, Federico Pasqualotto, Sergei Gukov, Jonathan N. Lee, Junsu Kim, 13 Kaiying Hou, Golnaz Ghiasi, Yi Tay, YaGuang Li, Chenkai Kuang, Yuan Liu, Hanzhao Lin, Evan Zheran Liu, Nigamaa Nayakanti, Xiaomeng Yang, Heng-Tze Cheng, Demi...

    arXiv 2026

  9. [9]

    Sign embedding quantum algorithms for matrix equations and matrix functions.arXiv preprint arXiv:2604.25333, 2026

    Yanqiao Wang and Jin-Peng Liu. Sign embedding quantum algorithms for matrix equations and matrix functions.arXiv preprint arXiv:2604.25333, 2026. URL https://arxiv.org/abs/2604.25333

    Pith/arXiv arXiv 2026

  10. [10]

    Generative language modeling for automated theorem proving.arXiv preprint arXiv:2009.03393, 2020

    Stanislas Polu and Ilya Sutskever. Generative language modeling for automated theorem proving.arXiv preprint arXiv:2009.03393, 2020. URL https://arxiv.org/abs/2009.03393

    Pith/arXiv arXiv 2009

  11. [11]

    HyperTree proof search for neural theorem proving

    Guillaume Lample, Timothée Lacroix, Marie-Anne Lachaux, Aurélien Rodriguez, Amaury Hayat, Thibaut Lavril, Gabriel Ebner, and Xavier Martinet. HyperTree proof search for neural theorem proving. InAdvances in Neural Information Processing Systems, volume 35, pages 26337–26349, 2022. URL https://proceedings. neurips.cc/paper_files/paper/2022/hash/a8901c5e85f...

    2022

  12. [12]

    The Lean theorem prover (system description)

    Leonardo de Moura, Soonho Kong, Jeremy Avigad, Floris van Doorn, and Jakob von Raumer. The Lean theorem prover (system description). InAutomated Deduction – CADE-25, volume 9195 ofLecture Notes in Computer Science, pages 378–388. Springer, 2015. doi: 10.1007/978-3-319-21401-6_26

    work page doi:10.1007/978-3-319-21401-6_26 2015

  13. [13]

    Towards end-to-end automation of ai research.Nature, 651(8107):914–919, March 2026

    Chris Lu, Cong Lu, Robert Tjarko Lange, Yutaro Yamada, Shengran Hu, Jakob Foerster, David Ha, and Jeff Clune. Towards end-to-end automation of AI research.Nature, 651(8107):914–919, 2026. doi: 10.1038/s41586-026-10265-5. URL https://doi.org/10.1038/s41586-026-10265-5

    work page doi:10.1038/s41586-026-10265-5 2026

  14. [14]

    Accelerating scientific discovery with Co- Scientist

    Juraj Gottweis, Wei-Hung Weng, Alexander Daryin, Tao Tu, Petar Sirkovic, Artiom Myaskovsky, Grzegorz Glowaty, Felix Weissenberger, Alessio Orlandi, Dan Popovici, et al. Accelerating scientific discovery with co- scientist.Nature,2026. doi: 10.1038/s41586-026-10644-y. URLhttps://doi.org/10.1038/s41586-026-10644-y. Published May 19, 2026

    work page doi:10.1038/s41586-026-10644-y 2026

  15. [15]

    E.; Chang, B.; Mitchener, L.; Yiu, A.; Szostkiewicz, C

    Ali Essam Ghareeb, Benjamin Chang, Ludovico Mitchener, Angela Yiu, Caralyn J. Szostkiewicz, Dmytro Shved, Gavin J. Gyimesi, Jon M. Laurent, Samantha M. Wright, Muhammed T. Razzak, Andrew D. White, Silvia C. Finnemann, Michaela M. Hinks, and Samuel G. Rodriques. A multi-agent system for automating scientific discovery.Nature, 2026. doi: 10.1038/s41586-026-...

    work page doi:10.1038/s41586-026-10652-y 2026

  16. [16]

    Advancingmathematicalresearchviahuman-AI interactive theorem proving.arXiv preprint arXiv:2512.09443, 2025

    ChenyiLi, ZhijianLai, DongAn, JiangHu, andZaiwenWen. Advancingmathematicalresearchviahuman-AI interactive theorem proving.arXiv preprint arXiv:2512.09443, 2025. URL https://arxiv.org/abs/2512.09443

    arXiv 2025

  17. [17]

    Daniel Zheng, Ingrid von Glehn, Yori Zwols, Iuliya Beloshapka, Lars Buesing, Daniel M. Roy, Martin Wattenberg, Bogdan Georgiev, Tatiana Schmidt, Andrew Cowie, Fernanda Viegas, Dimitri Kanevsky, Vineet Kahlon, Hartmut Maennel, Sophia Alj, George Holland, Alex Davies, and Pushmeet Kohli. AI co-mathematician: Accelerating mathematicians with agentic AI.arXiv...

    Pith/arXiv arXiv 2026

  18. [18]

    Harrow, Avinatan Hassidim, and Seth Lloyd

    Aram W. Harrow, Avinatan Hassidim, and Seth Lloyd. Quantum algorithm for linear systems of equations. Physical Review Letters, 103(15):150502, 2009. doi: 10.1103/PhysRevLett.103.150502

    work page doi:10.1103/physrevlett.103.150502 2009

  19. [19]

    and Kothari, Robin and Somma, Rolando D

    AndrewM.Childs,RobinKothari,andRolandoD.Somma. Quantumalgorithmforsystemsoflinearequations with exponentially improved dependence on precision.SIAM Journal on Computing, 46(6):1920–1950, 2017. doi: 10.1137/16M1087072

    work page internal anchor Pith review doi:10.1137/16m1087072 1920

  20. [20]

    Berry, Andrew M

    Dominic W. Berry, Andrew M. Childs, Richard Cleve, Robin Kothari, and Rolando D. Somma. Simulating hamiltonian dynamics with a truncated taylor series.Physical Review Letters, 114(9):090502, 2015. doi: 10.1103/PhysRevLett.114.090502

    work page doi:10.1103/physrevlett.114.090502 2015

  21. [21]

    Optimalhamiltoniansimulationbyquantumsignalprocessing.Physical Review Letters, 118(1):010501, 2017

    GuangHaoLowandIsaacL.Chuang. Optimalhamiltoniansimulationbyquantumsignalprocessing.Physical Review Letters, 118(1):010501, 2017. doi: 10.1103/PhysRevLett.118.010501

    work page doi:10.1103/physrevlett.118.010501 2017

  22. [22]

    Guang Hao Low and Isaac L. Chuang. Hamiltonian simulation by qubitization.Quantum, 3:163, 2019. doi: 10.22331/q-2019-07-12-163. 14

    work page doi:10.22331/q-2019-07-12-163 2019

  23. [23]

    Gily\' e n, Y

    András Gilyén, Yuan Su, Guang Hao Low, and Nathan Wiebe. Quantum singular value transformation and beyond: exponential improvements for quantum matrix arithmetics. InProceedings of the 51st Annual ACM SIGACT Symposium on Theory of Computing, pages 193–204. ACM, 2019. doi: 10.1145/3313276.3316366

    work page doi:10.1145/3313276.3316366 2019

  24. [24]

    Linear Combination of

    Dong An, Jin-Peng Liu, and Lin Lin. Linear Combination of Hamiltonian Simulation for Nonunitary Dynamics with Optimal State Preparation Cost.Physical Review Letters, 131(15):150603, 2023. doi: 10.1103/PhysRevLett.131.150603

    work page doi:10.1103/physrevlett.131.150603 2023

  25. [25]

    Childs, and Lin Lin

    Dong An, Andrew M. Childs, and Lin Lin. Quantum Algorithm for Linear Non-unitary Dynamics with Near-Optimal Dependence on All Parameters.Communications in Mathematical Physics, 407(1):19, 2026. doi: 10.1007/s00220-025-05509-w

    work page doi:10.1007/s00220-025-05509-w 2026

  26. [26]

    Childs, Lin Lin, and Lexing Ying

    Dong An, Andrew M. Childs, Lin Lin, and Lexing Ying. Laplace Transform–Based Quantum Eigenvalue Transformation via Linear Combination of Hamiltonian Simulation.SIAM Journal on Computing, 55(2): 376–409, 2026. doi: 10.1137/24M1720160

    work page doi:10.1137/24m1720160 2026

  27. [27]

    An efficient explicit implementation of a near-optimal quantum algorithm for simulating linear dissipative differential equations

    Ivan Novikau and Ilon Joseph. An Efficient Explicit Implementation of a Near-Optimal Quantum Algorithm for Simulating Linear Dissipative Differential Equations.arXiv preprint arXiv:2501.11146, 2025. doi: 10.48550/arXiv.2501.11146. URL https://arxiv.org/abs/2501.11146

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2501.11146 2025

  28. [28]

    Arbitrary Boundary Conditions and Constraints in Quantum Algorithms for Differential Equations via Penalty Projections

    Philipp Schleich, Tyler Kharazi, Xiangyu Li, Jin-Peng Liu, Alán Aspuru-Guzik, and Nathan Wiebe. Arbitrary BoundaryConditionsandConstraintsinQuantumAlgorithmsforDifferentialEquationsviaPenaltyProjections. arXiv preprint arXiv:2506.21751, 2025. doi: 10.48550/arXiv.2506.21751. URL https://arxiv.org/abs/2506. 21751

    work page doi:10.48550/arxiv.2506.21751 2025

  29. [29]

    doi: 10.48550/arXiv.2508.19238

    GuangHaoLowandRolandoD.Somma.OptimalQuantumSimulationofLinearNon-UnitaryDynamics.arXiv preprint arXiv:2508.19238, 2025. doi: 10.48550/arXiv.2508.19238. URL https://arxiv.org/abs/2508.19238

    work page doi:10.48550/arxiv.2508.19238 2025

  30. [30]

    Fourier Transform-Based Linear Combination of Hamiltonian Simulation.arXiv preprint arXiv:2508.19596, 2025

    Xi Huang and Dong An. Fourier Transform-Based Linear Combination of Hamiltonian Simulation.arXiv preprint arXiv:2508.19596, 2025. doi: 10.48550/arXiv.2508.19596. URL https://arxiv.org/abs/2508.19596

    work page doi:10.48550/arxiv.2508.19596 2025

  31. [31]

    Circuit-Efficient Randomized Quantum Simulation of Non-Unitary Dynamics with Observable-Driven and Symmetry-Aware Designs.arXiv preprint arXiv:2509.08030, 2025

    Songqinghao Yang and Jin-Peng Liu. Circuit-Efficient Randomized Quantum Simulation of Non-Unitary Dynamics with Observable-Driven and Symmetry-Aware Designs.arXiv preprint arXiv:2509.08030, 2025. doi: 10.48550/arXiv.2509.08030. URL https://arxiv.org/abs/2509.08030

    work page doi:10.48550/arxiv.2509.08030 2025

  32. [32]

    Quantum algorithms based on the Block-Encoding framework for matrix functions by contour integrals.Quantum Information and Computation, 22(11–12):965–979, 2022

    Souichi Takahira, Asuka Ohashi, Tomohiro Sogabe, and Tsuyoshi Sasaki Usuda. Quantum algorithms based on the Block-Encoding framework for matrix functions by contour integrals.Quantum Information and Computation, 22(11–12):965–979, 2022. doi: 10.26421/QIC22.11-12-4

    work page doi:10.26421/qic22.11-12-4 2022

  33. [33]

    Somma, Guang Hao Low, Dominic W

    Rolando D. Somma, Guang Hao Low, Dominic W. Berry, and Ryan Babbush. Quantum algorithm for linear matrix equations, 2025. arXiv:2508.02822

    arXiv 2025

  34. [34]

    Wilde, and Zhicheng Zhang

    Nana Liu, Qisheng Wang, Mark M. Wilde, and Zhicheng Zhang. Quantum algorithms for matrix geometric means.npj Quantum Information, 11:101, 2025. doi: 10.1038/s41534-025-00973-7

    work page doi:10.1038/s41534-025-00973-7 2025

  35. [35]

    doi:10.1137/1.9780898717778

    Nicholas J. Higham.Functions of Matrices: Theory and Computation. SIAM, Philadelphia, PA, 2008. doi: 10.1137/1.9780898717778

    work page doi:10.1137/1.9780898717778 2008

  36. [36]

    Kenney and Alan J

    Charles S. Kenney and Alan J. Laub. The matrix sign function.IEEE Transactions on Automatic Control, 40 (8):1330–1348, 1995. doi: 10.1109/9.402226

    work page doi:10.1109/9.402226 1995

  37. [37]

    Oxford University Press, Oxford, 1995

    Peter Lancaster and Leiba Rodman.Algebraic Riccati Equations. Oxford University Press, Oxford, 1995

    1995

  38. [38]

    J. D. Roberts. Linear model reduction and solution of the algebraic riccati equation by use of the sign function. International Journal of Control, 32(4):677–687, 1980. doi: 10.1080/00207178008922881

    work page doi:10.1080/00207178008922881 1980

  39. [39]

    Solving the algebraic riccati equation with the matrix sign function.Linear Algebra and Its Applications, 85:267–279, 1987

    Ralph Byers. Solving the algebraic riccati equation with the matrix sign function.Linear Algebra and Its Applications, 85:267–279, 1987. doi: 10.1016/0024-3795(87)90222-9. 15 A Experimental Results This section records representative interaction traces between the human researcher andAIMv2. Given a human input,AIMv2 first judges whether the problem is suf...

    work page doi:10.1016/0024-3795(87)90222-9 1987

  40. [40]

    the classical analytic representation that might be useful, such as rational approximations, resolvent formulas, contour integrals, matrix sign functions, or spectral projectors

  41. [41]

    the target quantum output model

  42. [42]

    the likely quantum primitives needed

  43. [43]

    what would be genuinely new compared with standard HHL/QSVT/contour-integral approaches

  44. [44]

    Do not try to force a single answer

    the main technical obstacles. Do not try to force a single answer. Give several candidate projects and rank them by conceptual unity, feasibility, and potential novelty. Assistant Entry (AIM’s Response) Below are candidate projects organized around the principle: InsteadofviewingQSVTonlyasawaytoimplementpolynomialapproximationstoscalarfunctions of spectra...

  45. [45]

    Instead of implementing𝑓 by a high-degree QSVT polynomial, implement a linear combination of shifted inverses

    Rational block-encodings of matrix functions via shifted resolvents Core idea Given a matrix function𝑓(𝐴), use a rational approximation 𝑓(𝐴) ≈𝑟(𝐴)=𝑐 0𝐼+ 𝑚∑︁ 𝑗=1 𝑐 𝑗 𝐴−𝑧 𝑗 𝐼 or, for Hermitian𝐴, 𝑓(𝐴) ≈𝑐 0𝐼+ 𝑚∑︁ 𝑗=1 𝑐 𝑗 (𝐴−𝑧 𝑗 𝐼) −1. Instead of implementing𝑓 by a high-degree QSVT polynomial, implement a linear combination of shifted inverses. 1.1. Classical ...

  46. [46]

    For example, if𝐴is stable, 𝑋= ∫ ∞ 0 𝑒𝑡 𝐴𝑄𝑒 𝑡 𝐴† 𝑑𝑡

    Quantum algorithms for Lyapunov and Sylvester equations using rational Krylov ideas Core idea Solve matrix equations such as 𝐴𝑋+𝑋 𝐵=𝐶, or the continuous Lyapunov equation 𝐴𝑋+𝑋 𝐴 † =−𝑄, 18 not by vectorizing and applying HHL to the huge linear system, but by using integral or rational representations of the solution operator. For example, if𝐴is stable, 𝑋= ...

  47. [47]

    Rational block-encodings of𝑓(𝐴)

  48. [48]

    Rational approximations for fractional powers

  49. [49]

    Matrix logarithm and inverse square root

  50. [50]

    Compiler for rational approximations. Main deliverable: A theorem converting a rational approximation 𝑟(𝑥)=𝑐 0 + 𝑚∑︁ 𝑗=1 𝑐 𝑗 𝑥−𝑧 𝑗 into a block-encoding of𝑟(𝐴) , with complexity depending explicitly on pole locations, coefficient ℓ1-norm, shifted condition numbers, and target error. This would give a rigorous backbone for the broader program. 20 Cluster B...

  51. [51]

    Spectral projectors via rational filters

  52. [52]

    Matrix sign and stable subspaces

  53. [53]

    Quantum FEAST-style algorithms

  54. [54]

    Green’s function and spectral density operators. Main deliverable: A quantum rational-filter framework that outputs block-encodings of spectral projectors or approxi- mate projectors, with complexity competitive against polynomial QSVT in small-gap regimes. This cluster is conceptually clean and application-rich. Cluster C: Matrix equations as rational op...

  55. [55]

    Lyapunov/Sylvester equations

  56. [56]

    ADI-inspired quantum algorithms

  57. [57]

    Bivariate separable rational approximations

  58. [58]

    Riccati equations. Main deliverable: Quantum algorithms for structured matrix equations that avoid black-box vectorization and instead exploit separable rational approximations of kernels such as 1 𝜆+𝜇 . This is probably the most novel cluster, but also technically harder. Cluster D: Rational preconditioners and reduced operators Projects:

  59. [59]

    Quantum rational preconditioners

  60. [60]

    Effective Hamiltonians

  61. [61]

    Main deliverable: Reusable block-encoded approximate inverses, Schur complements, or effective operators for downstream quantum algorithms

    Domain decomposition. Main deliverable: Reusable block-encoded approximate inverses, Schur complements, or effective operators for downstream quantum algorithms. This is potentially very impactful if one can define a convincing reuse model. Most promising individual projects Project 1: Rational block-encoding theorem for Stieltjes functions Focus on funct...

  62. [62]

    What is the correct quantum cost model for rational approximants? Classically, one minimizes approximation error or number of poles. Quantumly, the cost should include: Cost(𝑟) ≈ ∑︁ 𝑗 |𝑐 𝑗 | ·Cost (𝐴−𝑧 𝑗 𝐼) −1 or, more accurately, an LCU-normalized version involving 𝛼𝑟 =|𝑐 0| + ∑︁ 𝑗 |𝑐 𝑗 |𝛼 𝑗 , where𝛼 𝑗 is the block-encoding normalization of(𝐴−𝑧𝑗 𝐼) −1. T...

  63. [63]

    Quantumly, poles near the spectrum are expensive because shifted systems become ill-conditioned

    Can pole locations be optimized for quantum implementation cost? Classical pole optimization focuses on uniform approximation. Quantumly, poles near the spectrum are expensive because shifted systems become ill-conditioned. So one wants to solve a constrained rational approximation problem: min 𝑐 𝑗 ,𝑧 𝑗 [approximation error+𝜆·quantum implementation cost] ...

  64. [64]

    Rational methods are compelling only if the savings in approximation order survive the cost of shifted inverses

    When do rational methods beat polynomial QSVT? For Hermitian 𝐴, QSVT can implement many functions nearly optimally in polynomial degree. Rational methods are compelling only if the savings in approximation order survive the cost of shifted inverses. Promising regimes: •functions with poles or branch points near the spectral domain; •spectral projectors wi...

  65. [65]

    If𝐴is Hermitian, this can be reduced to real block systems such as 𝐴−𝑎𝐼 𝑏𝐼 −𝑏𝐼 𝐴−𝑎𝐼 −1

    How should one handle complex shifts? For𝑧=𝑎+𝑖𝑏, one needs to implement (𝐴−𝑧𝐼) −1. If𝐴is Hermitian, this can be reduced to real block systems such as 𝐴−𝑎𝐼 𝑏𝐼 −𝑏𝐼 𝐴−𝑎𝐼 −1 . Alternatively, one can pair conjugate poles to maintain Hermiticity: 𝑐(𝐴−𝑧𝐼) −1 + 𝑐(𝐴− 𝑧𝐼) −1. A good project would formalize efficient block-encodings for conjugate-paired rational terms

  66. [66]

    rational approximation to block-encoding

    What output model is most useful? Possible output models differ substantially: 1.Block-encodingof𝑓(𝐴): reusable and composable. 2.Stateproportional to𝑓(𝐴) |𝑏⟩: standard HHL-like model. 3.Density operatorproportional to𝑓(𝐴): useful when𝑓(𝐴) ⪰0. 4.Projector/reflection: useful for subspace algorithms. 5.Quantum channelrepresenting a solution map𝐶↦→𝑋. 6.Low-r...

  67. [67]

    the input access model, such as block-encodings of the coefficient matrices

  68. [68]

    the augmented matrix and the sign or projector identity

  69. [69]

    the rational or contour-integral approximation and its deterministic error

  70. [70]

    the quantum implementation of the shifted resolvents

  71. [71]

    the LCU normalization or block-encoding normalization

  72. [72]

    the total query complexity

  73. [73]

    explicitly implementable contractions with costs,

    how deterministic approximation error and quantum implementation error are combined. Finally,comparethisframeworkwithexistingquantumalgorithmsforSylvesterequations,Lyapunovequations, matrix functions, and contour-integral-based matrix-function algorithms. Identify regimes where the sign- embedding/rational-resolvent approach gives a genuine structural or ...